A Blog by Jonathan Low


Mar 17, 2018

Working In Tech: You Get What You Incent

Has the smart engineer-focused open plan, no dress code, foosball table, beer on Fridays model finally run its course - or simply seen its advantages degraded by the co-evolutionary reality that almost everyone who can has copied it? JL

Tyler Cowen reports in Marginal Revolution:

Trying to be in a business where there’s deep domain knowledge required, or lots of actual business competition (where marketing, awareness, and business strategy are key) mean that overfocus on really smart software engineers as the exclusive hiring target makes it difficult to succeed. The primary contributions the tech companies make to management are:decreasing the power managers hold (and) treating organization problems as systems problems to be designed, measured, optimized, and debugged
This is from my email, I have done a bit of minor editing to remove identifiers.  It is long, so it goes under the screen break:
I joined Google [earlier]…as an Engineering Director. This was, as I understand it, soon after an event where Larry either suggested or tried to fire all of the managers, believing they didn’t do much that was productive. (I’d say it was apocryphal but it did get written up in a Doc that had a bunch of Google lore, so it got enough oversight that it was probably at least somewhat accurate.)
At that time people were hammering on the doors trying to get in and some reasonably large subset, carefully vetted with stringent “smart tests” were being let in. The official mantra was, “hire the smartest people and they’ll figure out the right thing to do.” People were generally allowed to sign up for any project that interested them (there was a database where engineers could literally add your name to a project that interested you) and there was quite a bit of encouragement for people to relocate to remote offices. Someone (not Eric, I think it probably was Sergey) proposed opening offices anyplace there were smart people so that we could vacuum them up. Almost anything would be considered as a new project unless it was considered to be “not ambitious enough.” The food was fabulous. Recruiters, reportedly, told people they could work on “anything they wanted to.” There were microkitchens stocked with fabulous treats every 500′ and the toilets were fancy Japanese…uh…auto cleaning and drying types.
And… infrastructure projects and unglamorous projects went wanting for people to work on them. They had a half day meeting to review file system projects because…it turns out that many, many top computer scientists evidently dream of writing their own file systems. The level of entitlement displayed around things like which treats were provided at the microkitchens was…intense. (Later, there was a tragicomic story of when they changed bus schedules so that people couldn’t exploit the kitchens by getting meals for themselves [and family…seen that with my own eyes!] “to go” and take them home with them on the Google Bus — someone actually complained in a company meeting that the new schedules…meant they couldn’t get their meals to go. And they changed the bus schedule back, even though their intent was to reduce the abuse of the free food.)
Now, most of all that came from two sources not exclusively related to the question at hand:
Google (largely Larry I think) was fearless about trying new things. There was a general notion that we were so smart we could figure out a new, better way to do anything. That was really awesome. I’d say, overall, that it mostly didn’t pan out…but it did once in a while and it may well be that just thinking that way made working there so much fun, that it did make an atmosphere where, overall, great things happened.
Google was awash in money and happy to spray it all over its employees. Also awesome, but not something you can generalize for all businesses. Amazon, of course, took a very different tack. (It’s pretty painful to hear the stories in The Everything Store or similar books about the relatively Spartan conditions Amazon maintained. I was the site lead for the Google [xxxx] office for a while and we hired a fair number of Amazon refugees. They were really happy to be in Google, generally…not necessarily to either of our benefit.)
I was there for over ten years. Over time, the general rule of “you get what you incent” made the whole machine move much less well and the burdens of maintaining growth for Wall Street have had some real negative impact (Larry and Sergey have been pushing valiantly for some other big hit of course).
So, onto the question at hand:
I know bits and pieces about Google, Facebook, Apple, and Amazon. I’ve known some people who’ve worked at Netflix but generally know less about them. Google I know pretty well. I’ve worked at a bunch of startups and some bigger companies. I haven’t worked for a non-tech company (Ford) since I was 19 (when I was an undergrad I worked in the group that did the early engine control computers…a story in itself).
I think the primary contributions the tech companies make to organizational management are:
significantly decreasing the power that managers hold
treating organization problems as systems problems to be designed, measured, optimized, and debugged [as a manager, I, personally, treat human and emotional problems that way also]
high emphasis on employing top talent and very generous rewards distributed through the company*
*only possible in certain configurations of course.
What also went well at Google: Google avoided job categories that were, generally, likely to decrease accountability:
Google avoided the job class of architect — which was both high status and low accountability, making it an easy place for pricey senior people to park and not have much impact (Sun Microsystems was notorious for having lots and lots of architects)
Google avoided the category of project manager, which would have allowed engineering managers to avoid the grungy part of their job (and be out of touch with engineering realities). I don’t know the history of that particular orientation — we did have something called a TPM (“technical program manager”) who were intended to make deep technical contributions, not just keep track of projects.
Google exploited “level of indirection” to avoid giving managers power over their employees or the employees excess emotional bonds to their managers.
hiring committees who would remove the managers from the process of hiring and (mostly, especially in the early days) project assignment
promotion committees who would judge promotion cases, removing the power of promotion from the manager (didn’t scale well, as indicated by the link I sent you)
raises had a strong algorithmic component; promotions and bonuses were both linked to performance ratings in a way such that getting high scores (at the current level) led to big bonuses, so if an employee’s case wasn’t perfect for promotion they wouldn’t feel they were incurring a financial penalty. That gave promotion committees more liberty to say “no by default” and managers less incentive to fight like badgers to get their people promoted.
What didn’t go so well
The industry has its own weird relationship to business:
product managers can be valuable if they have either strong business skills or a deep instinct for something amazing that should be built to create a business. Google (and others) explicitly treated product managers as “mini-CEOs” so they attracted a lot of people who…wanted to be a mini-CEO…but weren’t necessarily cut out for a CEO role. (At this point I have a generally low opinion of product managers and people who aspire to product management, with notable exceptions of course.)
Google- and software industry-specific: lots of developers want to make free software, lots of developers only know how to make things for other developers, so trying to be in a business where there’s deep domain knowledge required, or lots of actual business competition (where marketing, awareness, and business strategy are key) mean that overfocus on really, really smart software engineers as the almost exclusive hiring target makes it difficult to succeed.
Selling ads…I’m not in favor of it as an engine of commerce. Amazon has profound and distinguished power accrued over time by ruthless exploitation of scale in low margin industries where everyone is “making it for a dollar, selling it for two…” which makes them very dangerous for every competitor.
You get what you incent
product managers were rewarded for launching, which means they’d tend to launch and ditch
it’s hard not to reward managers for group size; Google was no different — this was the place where it was hardest to avoid fiefdoms that come with centralization of power
What degraded over time at Google:
Some things having to do with too much money, not necessarily related to tech management in particular:
sense of company mission vs. sense of entitlement.
pursuing company mission vs. individual advancement.
influx of people responding primarily to financial rewards (related).
Some things related to scale that might work better in an organization based on tight, interpersonal relationships (the opposite of the decreased manager power referenced above):
some processes implicitly dependent on people largely knowing one another or being one degree of separation apart (e.g., promotion)
the ability to reward creative, risky work; the ability to reward engineering work that had little visible outcome.
Other companies in bits and pieces
As indicated I’m very admiring of Amazon’s strategic approach and its business-first focus. Google did a lot of awesome stuff, but it had incalculable waste and missed opportunities because of the level of pampering and scattershot approach. If you want a real tech company model, I’d pick Amazon (even though I’m not sure I’d ever work there).
Facebook is kind of nothing. It’s a product company and I (personally) don’t think the product is very compelling. I think they hit a moment and will see the fate of MySpace in time. I can’t pick out product innovations that were particularly awesome (other than incubating on college campuses and exploiting sex more or less tastefully). And, their infrastructure is pretty crude which means they’ll run into the problem, eventually, hiring the kind of people who can do the kind of scaling they’re going to need.
Apple — I don’t know a ton about them currently, but they’re old. Real old. I interviewed there some time ago and they told me they like to set arbitrary deadlines for their projects because once people are late they work harder. I didn’t pursue the job further, although I have no idea if that’s any sort of a broad practice or a current practice. What they *do* epitomize is the notion that new business models are more important than new technologies so things like flat rate data plans, $.99 songs, not licensing their OS, are real, interesting tech company contributions — I haven’t seen much of that sort of thing since Steve Jobs died, but I’m also not that close to them. That’s obviously not exclusive to tech companies, but something that may be more possible where you have new inventions.
Microsoft — the epitome of high pressure big software, abuse of market dominance, decline, and then pivot into new relevance. IBM II. I don’t know that there’s much about their culture or current business that’s particularly admirable. They’ve got this “partner” system that’s insane where they’ve set up a high stakes internal competition that just looks terrible for any kind of team cohesion or morale. I wouldn’t want to work there, either, although (like Amazon) I have a number of friends I really respect who work there. Generally, there are tradeoffs for having an environment with lots of competition for material rewards — I don’t personally like them so they won’t attract people like me… so I’d like to believe they’re terrible for business…although I’m not at all sure that’s true.
Netflix — little info, really. Competent and pivoting but I don’t know much good or bad.
Amazon — totally admirable, really scary, really effective, and very business-focused. Changing capex into opex via Cloud was one of those changes in business mode that I saw in Apple, along with “sell close to cost using Wall Street money so that no one can compete while you push down costs via scale so no one new can afford to enter the market.” They also are willing to ditch products that don’t work. It sounds like a hard place to work.
Challenges I see in other industries: low imagination, fiefdoms / politics, inefficiency, communication problems…all could benefit from tech company input. If you’re in a low margin, low revenue business…it’s just going to be hard without the ability to attract and retain top talent, which is usually going to have a money component. But, best practices certainly help along with awareness of the importance of things like business model, systems design within the business, communication and culture, relationships to power, politics, and incentives…
Remaining challenges in tech industry: scaling and incentives (and incentives at scale :). I also see a major extrovert bias, which might seem a little funny for tech. But, again, product managers (or, God forbid, Sales people) are all really subject to the “let’s just get some people in a room” style of planning and problem resolution. I firmly believe some massive amount of productivity is squandered from people choosing the wrong communication paradigm — I think it’s often chosen for the convenience or advantage of someone who is either in an extrovert role or who is just following extrovert tendencies. Massive problem at Google, which is ironic given their composition. Amazon had some obvious nods to avoiding these sorts of things (e.g., “reading time”) but I don’t know how pervasive they were or how effective people believed them to be.
I thank the author for taking the time to do this, of course I am presenting this content, not endorsing it.


Post a Comment